Re: [AMBER] many initial systems for a two molecules system

From: Jason Swails <>
Date: Thu, 4 Sep 2014 07:29:05 -0400

On Thu, Sep 4, 2014 at 3:44 AM, Fabian Glaser <>

> Hi Jason,
> Thanks a lot for your quick answer, my objective is to study the
> interaction between tow small molecules, but since I have no idea of how
> they interact, what I thought is to use implicit solvent model and then
> pull and push the molecules from each other slowly and keep 50 or 100
> conformations when they are close, to start a long dynamics for each of
> them. The idea is to obtain the largest possible sampling, without
> investing 3 months.The problem with the umbrella sampling (which I may not
> to understand correctly) is that I don't even know what is the appropriate
> distance between center of masses of both molecules, or when I should stop
> the simulation.

‚ÄčI wasn't suggesting actually doing umbrella sampling. I was suggesting
that you sample at fixed distances rather than do the constant
pushing/pulling. I believe that sampling at fixed distances will be a
better approach (it won't take as long and it'll give you more, better
starting spots at different distances for longer simulations).

However, I don't think that your approach is likely to yield interesting
results. Small molecules that are actually soluble rarely interact
strongly enough to stay close to each other under the conditions of
infinite dilution (and an implicit solvent simulation is infinitely
dilute). You are probably more likely to extract meaningful interaction
profiles by loading a solvent box with a handful of these compounds and
computing radial distribution functions (RDF). These are directly related
to the PMF along the RDF coordinate (namely, the distance between the two
species). The free energy obviously has a concentration dependence to it,
but you will know what that concentration is.


Jason M. Swails
Rutgers University
Postdoctoral Researcher
AMBER mailing list
Received on Thu Sep 04 2014 - 04:30:04 PDT
Custom Search